- Email: [email protected]
Reliability study of dye-sensitized solar cells by means of solar simulator and white LED
Microelectronics Reliability 52 (2012) 2495–2499
Contents lists available at SciVerse ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Reliability study of dye-sensitized solar cells by means of solar simulator and white LED D. Bari a,⇑, N. Wrachien a, R. Tagliaferro b, T.M. Brown b, A. Reale b, A. Di Carlo b, G. Meneghesso a, A. Cester a a b
DEI, Department of Information Engineering, University of Padova, Padova, Italy CHOSE, Department of Electronic Engineering, University of Rome ‘‘Tor Vergata’’, Roma, Italy
a r t i c l e
i n f o
Article history: Received 5 June 2012 Accepted 20 June 2012 Available online 11 July 2012
a b s t r a c t In this work, we take into account a LED-based light source as an alternative to AM1.5 solar simulator to perform the reliability study on dye-sensitized solar cells (DSCs). We performed accelerated optical stress by means of high power white LED and during stress we performed DC and EIS measurements with both white LED and AM1.5 solar simulator in order to find, if any, differences in kinetics degradation. During stress we also performed characterization measurements using monochromatic LED sources in order to understand if it adds more information about the DSCs degradation mechanism. We found that DC parameters feature different degradation rates depending on characterization source and differences also appear on degradation kinetics shape. The DSC characterization performed with monochromatic light sources show strong differences in degradation rate and in degradation kinetics shape as well depending on wavelength sources. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Suitability to many applications, low cost production, and low cost materials make dye-sensitized solar cells (DSCs) a new power generation device which is a promising alternative to conventional solar cells [1–3]. In spite of a promising role in the future of renewable energy market, the reliability of DSCs is still an open issue. Conventional solar cells certification can be made by means of Xenon lamp-based solar simulators which easy achieve the IEC Standards [4]. Nevertheless, this kind of solar simulators are expensive, difficult to handle and optical intensity regulation could shorten the lamp lifetime. Many works in literature deal with the construction of LED-based solar simulators, which are cheap and easy to handle [5]. Several works show that I–V curves measured by using LED-based simulators can be converted to STC I–V characteristics by means of IEC Standard correction methods. In addition, many laboratory tests, like high power inverter for solar panels testing, are independent of illumination sources and LED-based solar simulators can be used instead of expensive AM1.5 solar simulators. Even if these new solar simulators are more and more used in many applications, few works deal with the suitability of LEDbased solar simulators to perform characterization and reliability study of DSCs. In this work, we performed DSC ageing test by means of high power white LED and we investigate the suitability ⇑ Corresponding author. E-mail address: [email protected] (D. Bari). 0026-2714/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.microrel.2012.06.061
of LED-based solar simulators to perform the reliability study of DCSs by comparing the degradation kinetics measured with both AM 1.5 solar simulator and white LEDs. During stress, we also performed characterization measurements using four monochromatic light sources, i.e. red, amber, blue, and green lights in order to understand if different light source characterizations could give additional information on the degradation mechanism involved on DSCs.
2. Devices and experimental setup In this work, we considered ruthenium-based DSCs, fabricated by CHOSE Laboratory at the University of Rome ‘‘Tor Vergata’’. A prototype of this device is shown in Fig. 1. The core of each cell is the TiO2 at which is anchored the sensitized material. The 12 lm-TiO2 layer is deposited over one of the conductive TCO-glass substrate by sintering nanoparticles of TiO2 at high temperature. In order to improve the cell performances, an additional TiO2 scattering layer is deposited on the previous TiO2 layer. The ruthenium-based dye (N719) is impregnated in the nanoporous semiconductor. The resulting active area is 25 mm2. Using a thermoplastic gasket, the working electrode is hot-sealed with the second conductive Platinum-coated TCO-glass substrate which works as the counter electrode. Finally, the iodine/iodide-based liquid electrolyte is injected in the structure through the counter electrode by vacuum back filling technique. The pinhole is finally
2496
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
8
T=1h
AM1.5
Fresh
Current Density [mA/cm²]
7 6 5 T=0h T=1h T=4h T=26h T=50h T=91h T=155h T=343h T=671h T=1148h
4 3 2 1
Fig. 1. A sample with three embedded DSCs.
T=1148h
0 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Voltage [V]
(b)
12
Current Density [mA/cm²]
sealed by the same thermoplastic sealant previously used [6]. Each sample contains three electrically isolated cells as shown in Fig. 1. The experimental procedure begins with the characterization of the fresh devices. The characterization is performed at room temperature, inside a metallic box, which ensures electrostatic shielding and it prevents any disturbs from the environment. The characterization consists of: determination of the open-circuit voltage (VOC) and short-circuit current (ISC), the current density– voltage curves (J–V) measurements, and the electrochemical impedance spectroscopy analysis (EIS). Electrical characterization was performed with six different light sources: AM1.5 solar simulator, white LED and a RGBA LED. Multicolor LED has wavelengths of 630, 525, 460, and 590 nm and hereafter, we will refer to these wavelengths with ki, where i can be: R (red), G (green), B (blue), and A (amber) respectively. In order to compare the characterization performed with the six different sources, we carried out the characterization at same optical intensity of 60 mW/cm2 for all sources. We also measured the external and internal quantum efficiencies (EQE and IQE) for all ki. EQE (ki) is the ratio of the ISC measured under a monochromatic source (ki) divided by the number of incident photon per second. IQE (ki) was determined by dividing the EQE (ki) value by the absorption coefficient of the cell-stack at ki. After the characterization of the fresh devices, cells were illuminated at a constant optical intensity of 500 mW/cm2. During stress, the cells were loaded with a 12-X resistor, which keeps the cell close to short-circuit operating point. The stresses were periodically stopped to permit DC and EIS measurements. Due to the many light sources, in order to speed up the measurements, we used a custom automated measurement setup which, in addition, ensures good and stable electrical contacts between the devices and the instruments.
10
Fresh
White LED
8 6 4 2 0 0.0
T=0h T=1h T=4h T=26h T=50h T=91h T=155h T=343h T=671h T=1148h 0.1
T=1148h
0.2
0.3
0.4
0.5
0.6
0.7
Voltage [V] Fig. 2. Evolution of: current–density vs. voltage measured by means of AM1.5 solar simulator (a) and white LED (b).
components (absorption coefficients are 0.989 for both sources), while it partially absorbs amber and red light (absorption coefficients are 0.823 and 0.609 respectively). Figs. 5 and 6 show the evolution during stress of the EQE and IQE respectively. Fig. 7 shows the EIS characteristics: Fig. 7a and b shows the evolution of EIS plot measured by means of AM1.5 solar simulator and white LED respectively. In the next session the data collected during stress are discussed.
3. Results
4. Discussions
Figs. 2–4 show the data collected during optical stresses. Fig. 2a and b shows the current density–voltage (J–V) characteristic measured during stress by means of AM1.5 solar simulator and white LED respectively. Fig. 3a and b shows the degradation of the normalized electrical parameters JSC, and VOC as function of the stress time for all characterization light sources. The normalization has been done respected to the parameter values of the fresh device (see Table 1). Fig. 4 shows the emission spectrum density of the RGBA LEDs (continuous lines) and of White-LED (square dotted line). As comparison, in Fig. 4 is reported the cell-stack absorption spectrum (point dot line). As shown in figure, the white-LED emission spectrum partially covers the absorption spectrum of the sensitizer, in fact the dye spectrum has a tail in the near-IR and it has an absorption coefficient which still roughly one in the near-UV (up to 350 nm). Concerning the RGBA LED, the figure shows that dye material almost completely absorbs blue and green
Observing the data collected in Figs. 2 and 3 we can draw some considerations. The J–V characteristics measured by means of white LED and AM 1.5 solar simulator show slightly different evolutions. They feature a global decrease of the performance of the cell, i.e. a JSC decrease and a VOC reduction. Both measurements do not feature a variation in shape during stress. In particular, the curves appear parallel near the short-circuit and the open-circuit regions. This indicates that no remarkable variation of the series resistance (RS) and the shunt resistance (RP) occurs during the entire stress at least with an optical stress intensity of 5 Sun. At the same cell bias condition, some previous ageing tests carried out by means of AM 1.5 solar simulator have shown an increase of the JSC at the beginning of stress, in particular for optical intensities ranging from 500 mW/cm2 and 1000 mW/cm2 [7]. During white LED stress, the J–V characteristics measured by means of AM 1.5 solar simulator show a slight increment of the JSC within 1-h stress,
2497
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499 Table 1 Parameter values of fresh device used for normalization.
(a) 105 100
Param
Normalized JSC [%]
95
2
JSC (mA/cm ) VOC (mV) Efficiency (%)
90
80
Amber Blue Green Red White-Led AM.1.5
75
65 60 1
10
100
1000
Stress Time [h]
(b) 101 100 99
Normalized VOC [%]
White
R
G
B
A
6.8 708 3.32
10.6 720 4.89
9.1 718 4.36
9.2 715 4.31
8.9 712 4.17
8 711 3.88
85
70
98 97 96 95 94 93 92 91
Amber Blue Green Red White-Led AM.1.5
90 10
100
1000
Stress Time [h] Fig. 3. Evolution of: short-circuit current density vs. time (a), open-circuit voltage vs. time (b).
0n m 63
1.0
52 5n m
1.1
59 0n m
1.2
45 0 46 nm 0n m
Normalized Power Spectral Density [a.u.]
AM1.5
Red Amber Green Blue White-LED Absorption Spectrum
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 400
450
500
550
600
650
700
750
Wavelength [nm] Fig. 4. RGBA LED, white-LED emission spectrum and cell-stack absorption spectrum.
while the J–V measured by means of white LED does not show the same initial increment. Two considerations are worth to be drawn. Firstly, the JSC increment observed during AM 1.5 solar simulator stresses can be ascribed to an annealing phase, likely due to the wide presence of near-IR component in the solar simulator spectra,
which might induce a moderate heating of cell materials. Since the IR component is not present in white LED spectrum (see Fig. 4), no significant annealing can occur. The slight increase of JSC observed during white LED stress, can be ascribed to self-heating due to a combination of light exposure and electrical conduction. Secondly, even though the JSC increment is marginal, the capability of AM 1.5 solar simulator to detect it, points out that white LED could not be able to show the complete kinetic degradation J–V characteristics, regardless of the stress source. From the J–V and efficiency–V curves measured by means of RGBA LED, only green LED show J–V evolution similar to white LED. In fact, green light is a main component in white LEDs besides blue light. Efficiency–V curves show the same evolution of the J–V curves. Fig. 3a and b show the evolution of JSC and VOC for the six light sources employed. Fig. 3b shows a monotonic decrease of VOC as stress time increases, regardless of the characterization sources. White LED and AM 1.5 solar simulator light sources show similar degradation rate giving the same information on the degradation kinetic of VOC during stress, at least with an optical stress intensity of 5 Sun. The degradation kinetic features almost the same shape for all light sources, but with different degradation rates. In fact, the characterization RGBA LED show a minor reduction of VOC compared with the reduction observed with solar simulator and white led light sources. Fig. 3a shows the evolution of the JSC during stress for all six characterization light sources. The JSC kinetic measured by means of AM 1.5 solar simulator and white LED show a 10% and 2% drop of the JSC respectively within 10 h of stress. Ruthenium-based dyes absorb radiation ranging from 300–700 nm and standard AM 1.5 solar simulator spectrum widely covers this wavelength range. Tentatively, we suppose that white light exposure combined with electrical conduction induces dye degradation and, in particular, making the dye no longer able to absorb wavelength in the near-UV (300–400 nm). Since white LED partially covers the absorption spectrum of the dye (the range spectra of a white led ranges from 400 to 700 nm, see Fig. 4), characterization by means of white LED cannot detect degradation that occurs outside of the white spectrum. JSC kinetics measured by means of RGBA LED show different evolutions. Characterization with red and blue light sources show an initial increase JSC within 50-h stress and then JSC monotonically decreases; amber and green light characterization show a monotonic decrease of JSC since the beginning of the stress. After about 1100-h of stress, JSC measured with red, amber, blue, and green lights shows a drop 19%, 23%, 26%, and 27% respectively. We can conclude that different wavelength light sources give different information on the degradation kinetics of DC parameters (i.e. VOC, JSC) of a solar cell. We also found that degradation kinetics of fill factor and efficiency follows the same kinetics of JSC. Figs. 5 and 6 show the evolution during stress of the EQE and IQE of the cell measured by different wavelength of the light source, in particular amber, blues, green, and red colors. As shown in Fig. 4, the dye material well absorbs light with a wavelength ranging from 400 to 550 nm, while it partially absorbs wavelength greater than 550 nm. Referring to the fresh curves in Fig. 5a–d (black line) and to a fixed photons rate, i.e. 5 1016 photons/s, EQE is higher for blue and green light (about 36% for both light) than for amber and red light (28% and 30% respectively). As comparison, in Fig. 6a–d is reported the IQE of the cell measured with
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
34 32
EQE [%]
30
(c) 42
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Amber Source
Fresh
28
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Green Source
40 38
26
Fresh
36
EQE [%]
2498
34 32 30 28
24 T=1148h
T=1148h
26
22 3x10
16
4x10
16
5x10
16
6x10
16
16
3x10
16
16
4x10
5x10
Photons [1/s] 42 38 36 34 32 30 28 26 24 22 20 18
(d)
Blue Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Fresh
16
7x10
16
8x10
16
9x10
Photons [1/s] 31
Red Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
30
Fresh
29 28
EQE [%]
EQE [%]
(b) 40
16
6x10
27 26 25 24 23
T=1148h
22
T=1148h
21 16
4.0x10
16
8.0x10
17
1.2x10
17
1.6x10
16
17
4.0x10
2.0x10
16
8.0x10
17
1.2x10
17
1.6x10
17
2.0x10
17
2.4x10
Photons [1/s]
Photons [1/s]
Fig. 5. EQE evolution measured by different light sources: amber (a), blue (b), green (c), and red (d). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
40
IQE [%]
38 36
(c)
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Amber Source
Fresh
34
42
38
32
Fresh
36 34 32 30
30
28
28
T=1148h
T=1148h 26
26
16
16
3x10
16
4x10
5x10
16
16
4.0x10
6x10
Photons [1/s] 42 40 38 36 34 32 30 28 26 24 22 20 18
Blue Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Fresh
16
6.0x10
16
8.0x10
17
1.0x10
Photons [1/s]
(d) 52 50 48
Fresh
46
IQE [%]
IQE [%]
(b)
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Green Source
40
IQE [%]
(a) 42
44 42
Red Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
40 38
T=1148h
36 4.0x10
16
8.0x10
16
1.2x10
17
1.6x10
Photons [1/s]
17
2.0x10
17
T=1148h
34 16 16 17 17 17 17 4.0x10 8.0x10 1.2x10 1.6x10 2.0x10 2.4x10
Photons [1/s]
Fig. 6. IQE evolution measured by different light sources: amber (a), blue (b), green (c), and red (d). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
18
T=0h T=10h T=26h T=91h T=155h T=343h T=671h T=1148h
16 14
-Z'' [ohm]
12 10
T=1148h AM 1.5
8
Fresh
6 4 2 0 25
30
35
40
45
50
55
60
65
70
75
80
85
Z' [ohm]
(b)
10
T=0h T=10h T=26h T=91h T=155h T=343h T=671h T=1148h
9 8
-Z'' [ohm]
7 6
White LED
In this work, we analyzed the suitability of white LED as stress and characterization light source. We found that exposure to white light leads to the degradation of DSC performances. In particular, we found that, even though white LED spectrum has not near-UV component, dye molecules are not longer be able to absorb wavelength ranging from 300–400 nm. On the other hand, among the six light sources, only AM 1.5 solar simulator pointed out the degradation of the dye absorption spectrum in the near-UV. Since variation in dye absorption spectrum is not uniform during optical and electrical stress, monochromatic light sources can be used to detect variation in tight absorption spectrum region giving more information on DSCs degradation mechanism.
5 Fresh
3 2 1 0 25
30
35
40
45
50
55
60
Fig. 7a–b shows the typical DSC Nyquist plot measured during stress by means of AM 1.5 solar simulator and white LED. EIS characteristics show an enlargement of the second semicircle, which is correlated to the TiO2/dye/electrolyte interface [8]. In agreement with JSC degradation kinetics, the degradation of this interface can be ascribed to dye molecules degradation, which are no longer be able to photo generate electrical charge. The enlargement shown during stress is regardless of the characterization source, nevertheless it is more pronounced if the characterization is performed with AM 1.5 light source. At the end of stress, the third semicircles, which is correlated to the Nernst diffusion in the electrolyte, does not show appreciable variation. The first semicircle, which is related to the back reaction at the counter electrode, features an enlargement regardless of the characterization source. EIS measurements carried out with monochromatic light sources also show an enlargement of the second semicircle, nevertheless they feature different enlargement variations. 5. Conclusions
T=1148h
4
2499
65
70
Z' [ohm] Fig. 7. Evolution of EIS characteristics measured by means of AM1.5 solar simulator (a) and white LED (b).
all colors. Even if the dye absorption coefficient for red light is lower than green and blue light, the IQE is higher for red light than blue and green light. This fact allows us to suppose that the extraction mechanism is more efficient for electron/hole pair photo-generated by red light than by blue or green light. At present, we are still investigating these phenomena. The evolutions during stress of the quantum efficiencies vs. incident photon rates characteristics depend on the characterization wavelength source. In fact, the efficiency reductions are different for the four colors. Referring to a constant incident photon rate of 5 1016 photons/s, we observed a reduction of 12%, 13%, 20%, and 22% of quantum efficiencies for amber, red, green, and blue characterization light sources at the end of stress. We suppose that during stress, the most degradation occurs for those binds of the dye, which are involved in the photo-generation of electron/hole pairs when green and blue light hit the sensitizer.
Acknowledgments This work was partially supported by progetto di Ateneo 2009 – Università di Padova, Italy (Project Number CPDA083941), progetto regione Veneto - SMUPR n. 4148 ‘‘Polo di ricerca nel settore Fotovoltaico’’ POR CRO PARTE FESR 2007–2013 Azione 1.1.1 a regia Regionale and by ENIAC JU/CALL 2010/270722-2 ERG ‘‘Energy for a green society: from sustainable harvesting to smart distribution equipments, materials, design solution and their applications’’. References [1] Gratzel M. Dye-sensitized solar cell. J Photochem Photobio A 2003;4:145–53. [2] O’Regan B et al. A low-cost, high efficiency solar cell based on dye-sensitized colloidal TiO2 films. Nature 1991;353:737–40. [3] Gratzel M. Nature 2001;414:338. [4] Standards, IEC 60904–9 photovoltaic devices – Part 9: solar simulator performance requirements, edition 2.0; 2007. p. 1–30. [5] Kohraku S, et al. New method for solar cell measurement by LED solar simulator. In: 3rd World conference on photovoltaic energy conference, Osaka, Japan; 11–18 May 2003. p. 1977–80. [6] Liberatore M et al. J Appl Electrochem 2009;39:2291–5. [7] Bari D, et al. IRPS; 2011. [8] Qing W et al. Electrochemical impedance spectroscopic analysis of dyesensitized solar cells. J Phys Chem B 2005;109:14945–53.
Contents lists available at SciVerse ScienceDirect
Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel
Reliability study of dye-sensitized solar cells by means of solar simulator and white LED D. Bari a,⇑, N. Wrachien a, R. Tagliaferro b, T.M. Brown b, A. Reale b, A. Di Carlo b, G. Meneghesso a, A. Cester a a b
DEI, Department of Information Engineering, University of Padova, Padova, Italy CHOSE, Department of Electronic Engineering, University of Rome ‘‘Tor Vergata’’, Roma, Italy
a r t i c l e
i n f o
Article history: Received 5 June 2012 Accepted 20 June 2012 Available online 11 July 2012
a b s t r a c t In this work, we take into account a LED-based light source as an alternative to AM1.5 solar simulator to perform the reliability study on dye-sensitized solar cells (DSCs). We performed accelerated optical stress by means of high power white LED and during stress we performed DC and EIS measurements with both white LED and AM1.5 solar simulator in order to find, if any, differences in kinetics degradation. During stress we also performed characterization measurements using monochromatic LED sources in order to understand if it adds more information about the DSCs degradation mechanism. We found that DC parameters feature different degradation rates depending on characterization source and differences also appear on degradation kinetics shape. The DSC characterization performed with monochromatic light sources show strong differences in degradation rate and in degradation kinetics shape as well depending on wavelength sources. Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction Suitability to many applications, low cost production, and low cost materials make dye-sensitized solar cells (DSCs) a new power generation device which is a promising alternative to conventional solar cells [1–3]. In spite of a promising role in the future of renewable energy market, the reliability of DSCs is still an open issue. Conventional solar cells certification can be made by means of Xenon lamp-based solar simulators which easy achieve the IEC Standards [4]. Nevertheless, this kind of solar simulators are expensive, difficult to handle and optical intensity regulation could shorten the lamp lifetime. Many works in literature deal with the construction of LED-based solar simulators, which are cheap and easy to handle [5]. Several works show that I–V curves measured by using LED-based simulators can be converted to STC I–V characteristics by means of IEC Standard correction methods. In addition, many laboratory tests, like high power inverter for solar panels testing, are independent of illumination sources and LED-based solar simulators can be used instead of expensive AM1.5 solar simulators. Even if these new solar simulators are more and more used in many applications, few works deal with the suitability of LEDbased solar simulators to perform characterization and reliability study of DSCs. In this work, we performed DSC ageing test by means of high power white LED and we investigate the suitability ⇑ Corresponding author. E-mail address: [email protected] (D. Bari). 0026-2714/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.microrel.2012.06.061
of LED-based solar simulators to perform the reliability study of DCSs by comparing the degradation kinetics measured with both AM 1.5 solar simulator and white LEDs. During stress, we also performed characterization measurements using four monochromatic light sources, i.e. red, amber, blue, and green lights in order to understand if different light source characterizations could give additional information on the degradation mechanism involved on DSCs.
2. Devices and experimental setup In this work, we considered ruthenium-based DSCs, fabricated by CHOSE Laboratory at the University of Rome ‘‘Tor Vergata’’. A prototype of this device is shown in Fig. 1. The core of each cell is the TiO2 at which is anchored the sensitized material. The 12 lm-TiO2 layer is deposited over one of the conductive TCO-glass substrate by sintering nanoparticles of TiO2 at high temperature. In order to improve the cell performances, an additional TiO2 scattering layer is deposited on the previous TiO2 layer. The ruthenium-based dye (N719) is impregnated in the nanoporous semiconductor. The resulting active area is 25 mm2. Using a thermoplastic gasket, the working electrode is hot-sealed with the second conductive Platinum-coated TCO-glass substrate which works as the counter electrode. Finally, the iodine/iodide-based liquid electrolyte is injected in the structure through the counter electrode by vacuum back filling technique. The pinhole is finally
2496
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
8
T=1h
AM1.5
Fresh
Current Density [mA/cm²]
7 6 5 T=0h T=1h T=4h T=26h T=50h T=91h T=155h T=343h T=671h T=1148h
4 3 2 1
Fig. 1. A sample with three embedded DSCs.
T=1148h
0 0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Voltage [V]
(b)
12
Current Density [mA/cm²]
sealed by the same thermoplastic sealant previously used [6]. Each sample contains three electrically isolated cells as shown in Fig. 1. The experimental procedure begins with the characterization of the fresh devices. The characterization is performed at room temperature, inside a metallic box, which ensures electrostatic shielding and it prevents any disturbs from the environment. The characterization consists of: determination of the open-circuit voltage (VOC) and short-circuit current (ISC), the current density– voltage curves (J–V) measurements, and the electrochemical impedance spectroscopy analysis (EIS). Electrical characterization was performed with six different light sources: AM1.5 solar simulator, white LED and a RGBA LED. Multicolor LED has wavelengths of 630, 525, 460, and 590 nm and hereafter, we will refer to these wavelengths with ki, where i can be: R (red), G (green), B (blue), and A (amber) respectively. In order to compare the characterization performed with the six different sources, we carried out the characterization at same optical intensity of 60 mW/cm2 for all sources. We also measured the external and internal quantum efficiencies (EQE and IQE) for all ki. EQE (ki) is the ratio of the ISC measured under a monochromatic source (ki) divided by the number of incident photon per second. IQE (ki) was determined by dividing the EQE (ki) value by the absorption coefficient of the cell-stack at ki. After the characterization of the fresh devices, cells were illuminated at a constant optical intensity of 500 mW/cm2. During stress, the cells were loaded with a 12-X resistor, which keeps the cell close to short-circuit operating point. The stresses were periodically stopped to permit DC and EIS measurements. Due to the many light sources, in order to speed up the measurements, we used a custom automated measurement setup which, in addition, ensures good and stable electrical contacts between the devices and the instruments.
10
Fresh
White LED
8 6 4 2 0 0.0
T=0h T=1h T=4h T=26h T=50h T=91h T=155h T=343h T=671h T=1148h 0.1
T=1148h
0.2
0.3
0.4
0.5
0.6
0.7
Voltage [V] Fig. 2. Evolution of: current–density vs. voltage measured by means of AM1.5 solar simulator (a) and white LED (b).
components (absorption coefficients are 0.989 for both sources), while it partially absorbs amber and red light (absorption coefficients are 0.823 and 0.609 respectively). Figs. 5 and 6 show the evolution during stress of the EQE and IQE respectively. Fig. 7 shows the EIS characteristics: Fig. 7a and b shows the evolution of EIS plot measured by means of AM1.5 solar simulator and white LED respectively. In the next session the data collected during stress are discussed.
3. Results
4. Discussions
Figs. 2–4 show the data collected during optical stresses. Fig. 2a and b shows the current density–voltage (J–V) characteristic measured during stress by means of AM1.5 solar simulator and white LED respectively. Fig. 3a and b shows the degradation of the normalized electrical parameters JSC, and VOC as function of the stress time for all characterization light sources. The normalization has been done respected to the parameter values of the fresh device (see Table 1). Fig. 4 shows the emission spectrum density of the RGBA LEDs (continuous lines) and of White-LED (square dotted line). As comparison, in Fig. 4 is reported the cell-stack absorption spectrum (point dot line). As shown in figure, the white-LED emission spectrum partially covers the absorption spectrum of the sensitizer, in fact the dye spectrum has a tail in the near-IR and it has an absorption coefficient which still roughly one in the near-UV (up to 350 nm). Concerning the RGBA LED, the figure shows that dye material almost completely absorbs blue and green
Observing the data collected in Figs. 2 and 3 we can draw some considerations. The J–V characteristics measured by means of white LED and AM 1.5 solar simulator show slightly different evolutions. They feature a global decrease of the performance of the cell, i.e. a JSC decrease and a VOC reduction. Both measurements do not feature a variation in shape during stress. In particular, the curves appear parallel near the short-circuit and the open-circuit regions. This indicates that no remarkable variation of the series resistance (RS) and the shunt resistance (RP) occurs during the entire stress at least with an optical stress intensity of 5 Sun. At the same cell bias condition, some previous ageing tests carried out by means of AM 1.5 solar simulator have shown an increase of the JSC at the beginning of stress, in particular for optical intensities ranging from 500 mW/cm2 and 1000 mW/cm2 [7]. During white LED stress, the J–V characteristics measured by means of AM 1.5 solar simulator show a slight increment of the JSC within 1-h stress,
2497
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499 Table 1 Parameter values of fresh device used for normalization.
(a) 105 100
Param
Normalized JSC [%]
95
2
JSC (mA/cm ) VOC (mV) Efficiency (%)
90
80
Amber Blue Green Red White-Led AM.1.5
75
65 60 1
10
100
1000
Stress Time [h]
(b) 101 100 99
Normalized VOC [%]
White
R
G
B
A
6.8 708 3.32
10.6 720 4.89
9.1 718 4.36
9.2 715 4.31
8.9 712 4.17
8 711 3.88
85
70
98 97 96 95 94 93 92 91
Amber Blue Green Red White-Led AM.1.5
90 10
100
1000
Stress Time [h] Fig. 3. Evolution of: short-circuit current density vs. time (a), open-circuit voltage vs. time (b).
0n m 63
1.0
52 5n m
1.1
59 0n m
1.2
45 0 46 nm 0n m
Normalized Power Spectral Density [a.u.]
AM1.5
Red Amber Green Blue White-LED Absorption Spectrum
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 400
450
500
550
600
650
700
750
Wavelength [nm] Fig. 4. RGBA LED, white-LED emission spectrum and cell-stack absorption spectrum.
while the J–V measured by means of white LED does not show the same initial increment. Two considerations are worth to be drawn. Firstly, the JSC increment observed during AM 1.5 solar simulator stresses can be ascribed to an annealing phase, likely due to the wide presence of near-IR component in the solar simulator spectra,
which might induce a moderate heating of cell materials. Since the IR component is not present in white LED spectrum (see Fig. 4), no significant annealing can occur. The slight increase of JSC observed during white LED stress, can be ascribed to self-heating due to a combination of light exposure and electrical conduction. Secondly, even though the JSC increment is marginal, the capability of AM 1.5 solar simulator to detect it, points out that white LED could not be able to show the complete kinetic degradation J–V characteristics, regardless of the stress source. From the J–V and efficiency–V curves measured by means of RGBA LED, only green LED show J–V evolution similar to white LED. In fact, green light is a main component in white LEDs besides blue light. Efficiency–V curves show the same evolution of the J–V curves. Fig. 3a and b show the evolution of JSC and VOC for the six light sources employed. Fig. 3b shows a monotonic decrease of VOC as stress time increases, regardless of the characterization sources. White LED and AM 1.5 solar simulator light sources show similar degradation rate giving the same information on the degradation kinetic of VOC during stress, at least with an optical stress intensity of 5 Sun. The degradation kinetic features almost the same shape for all light sources, but with different degradation rates. In fact, the characterization RGBA LED show a minor reduction of VOC compared with the reduction observed with solar simulator and white led light sources. Fig. 3a shows the evolution of the JSC during stress for all six characterization light sources. The JSC kinetic measured by means of AM 1.5 solar simulator and white LED show a 10% and 2% drop of the JSC respectively within 10 h of stress. Ruthenium-based dyes absorb radiation ranging from 300–700 nm and standard AM 1.5 solar simulator spectrum widely covers this wavelength range. Tentatively, we suppose that white light exposure combined with electrical conduction induces dye degradation and, in particular, making the dye no longer able to absorb wavelength in the near-UV (300–400 nm). Since white LED partially covers the absorption spectrum of the dye (the range spectra of a white led ranges from 400 to 700 nm, see Fig. 4), characterization by means of white LED cannot detect degradation that occurs outside of the white spectrum. JSC kinetics measured by means of RGBA LED show different evolutions. Characterization with red and blue light sources show an initial increase JSC within 50-h stress and then JSC monotonically decreases; amber and green light characterization show a monotonic decrease of JSC since the beginning of the stress. After about 1100-h of stress, JSC measured with red, amber, blue, and green lights shows a drop 19%, 23%, 26%, and 27% respectively. We can conclude that different wavelength light sources give different information on the degradation kinetics of DC parameters (i.e. VOC, JSC) of a solar cell. We also found that degradation kinetics of fill factor and efficiency follows the same kinetics of JSC. Figs. 5 and 6 show the evolution during stress of the EQE and IQE of the cell measured by different wavelength of the light source, in particular amber, blues, green, and red colors. As shown in Fig. 4, the dye material well absorbs light with a wavelength ranging from 400 to 550 nm, while it partially absorbs wavelength greater than 550 nm. Referring to the fresh curves in Fig. 5a–d (black line) and to a fixed photons rate, i.e. 5 1016 photons/s, EQE is higher for blue and green light (about 36% for both light) than for amber and red light (28% and 30% respectively). As comparison, in Fig. 6a–d is reported the IQE of the cell measured with
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
34 32
EQE [%]
30
(c) 42
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Amber Source
Fresh
28
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Green Source
40 38
26
Fresh
36
EQE [%]
2498
34 32 30 28
24 T=1148h
T=1148h
26
22 3x10
16
4x10
16
5x10
16
6x10
16
16
3x10
16
16
4x10
5x10
Photons [1/s] 42 38 36 34 32 30 28 26 24 22 20 18
(d)
Blue Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Fresh
16
7x10
16
8x10
16
9x10
Photons [1/s] 31
Red Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
30
Fresh
29 28
EQE [%]
EQE [%]
(b) 40
16
6x10
27 26 25 24 23
T=1148h
22
T=1148h
21 16
4.0x10
16
8.0x10
17
1.2x10
17
1.6x10
16
17
4.0x10
2.0x10
16
8.0x10
17
1.2x10
17
1.6x10
17
2.0x10
17
2.4x10
Photons [1/s]
Photons [1/s]
Fig. 5. EQE evolution measured by different light sources: amber (a), blue (b), green (c), and red (d). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
40
IQE [%]
38 36
(c)
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Amber Source
Fresh
34
42
38
32
Fresh
36 34 32 30
30
28
28
T=1148h
T=1148h 26
26
16
16
3x10
16
4x10
5x10
16
16
4.0x10
6x10
Photons [1/s] 42 40 38 36 34 32 30 28 26 24 22 20 18
Blue Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Fresh
16
6.0x10
16
8.0x10
17
1.0x10
Photons [1/s]
(d) 52 50 48
Fresh
46
IQE [%]
IQE [%]
(b)
T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
Green Source
40
IQE [%]
(a) 42
44 42
Red Source T=0h T=1h T=26h T=91h T=155h T=343h T=671h T=1148h
40 38
T=1148h
36 4.0x10
16
8.0x10
16
1.2x10
17
1.6x10
Photons [1/s]
17
2.0x10
17
T=1148h
34 16 16 17 17 17 17 4.0x10 8.0x10 1.2x10 1.6x10 2.0x10 2.4x10
Photons [1/s]
Fig. 6. IQE evolution measured by different light sources: amber (a), blue (b), green (c), and red (d). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
D. Bari et al. / Microelectronics Reliability 52 (2012) 2495–2499
(a)
18
T=0h T=10h T=26h T=91h T=155h T=343h T=671h T=1148h
16 14
-Z'' [ohm]
12 10
T=1148h AM 1.5
8
Fresh
6 4 2 0 25
30
35
40
45
50
55
60
65
70
75
80
85
Z' [ohm]
(b)
10
T=0h T=10h T=26h T=91h T=155h T=343h T=671h T=1148h
9 8
-Z'' [ohm]
7 6
White LED
In this work, we analyzed the suitability of white LED as stress and characterization light source. We found that exposure to white light leads to the degradation of DSC performances. In particular, we found that, even though white LED spectrum has not near-UV component, dye molecules are not longer be able to absorb wavelength ranging from 300–400 nm. On the other hand, among the six light sources, only AM 1.5 solar simulator pointed out the degradation of the dye absorption spectrum in the near-UV. Since variation in dye absorption spectrum is not uniform during optical and electrical stress, monochromatic light sources can be used to detect variation in tight absorption spectrum region giving more information on DSCs degradation mechanism.
5 Fresh
3 2 1 0 25
30
35
40
45
50
55
60
Fig. 7a–b shows the typical DSC Nyquist plot measured during stress by means of AM 1.5 solar simulator and white LED. EIS characteristics show an enlargement of the second semicircle, which is correlated to the TiO2/dye/electrolyte interface [8]. In agreement with JSC degradation kinetics, the degradation of this interface can be ascribed to dye molecules degradation, which are no longer be able to photo generate electrical charge. The enlargement shown during stress is regardless of the characterization source, nevertheless it is more pronounced if the characterization is performed with AM 1.5 light source. At the end of stress, the third semicircles, which is correlated to the Nernst diffusion in the electrolyte, does not show appreciable variation. The first semicircle, which is related to the back reaction at the counter electrode, features an enlargement regardless of the characterization source. EIS measurements carried out with monochromatic light sources also show an enlargement of the second semicircle, nevertheless they feature different enlargement variations. 5. Conclusions
T=1148h
4
2499
65
70
Z' [ohm] Fig. 7. Evolution of EIS characteristics measured by means of AM1.5 solar simulator (a) and white LED (b).
all colors. Even if the dye absorption coefficient for red light is lower than green and blue light, the IQE is higher for red light than blue and green light. This fact allows us to suppose that the extraction mechanism is more efficient for electron/hole pair photo-generated by red light than by blue or green light. At present, we are still investigating these phenomena. The evolutions during stress of the quantum efficiencies vs. incident photon rates characteristics depend on the characterization wavelength source. In fact, the efficiency reductions are different for the four colors. Referring to a constant incident photon rate of 5 1016 photons/s, we observed a reduction of 12%, 13%, 20%, and 22% of quantum efficiencies for amber, red, green, and blue characterization light sources at the end of stress. We suppose that during stress, the most degradation occurs for those binds of the dye, which are involved in the photo-generation of electron/hole pairs when green and blue light hit the sensitizer.
Acknowledgments This work was partially supported by progetto di Ateneo 2009 – Università di Padova, Italy (Project Number CPDA083941), progetto regione Veneto - SMUPR n. 4148 ‘‘Polo di ricerca nel settore Fotovoltaico’’ POR CRO PARTE FESR 2007–2013 Azione 1.1.1 a regia Regionale and by ENIAC JU/CALL 2010/270722-2 ERG ‘‘Energy for a green society: from sustainable harvesting to smart distribution equipments, materials, design solution and their applications’’. References [1] Gratzel M. Dye-sensitized solar cell. J Photochem Photobio A 2003;4:145–53. [2] O’Regan B et al. A low-cost, high efficiency solar cell based on dye-sensitized colloidal TiO2 films. Nature 1991;353:737–40. [3] Gratzel M. Nature 2001;414:338. [4] Standards, IEC 60904–9 photovoltaic devices – Part 9: solar simulator performance requirements, edition 2.0; 2007. p. 1–30. [5] Kohraku S, et al. New method for solar cell measurement by LED solar simulator. In: 3rd World conference on photovoltaic energy conference, Osaka, Japan; 11–18 May 2003. p. 1977–80. [6] Liberatore M et al. J Appl Electrochem 2009;39:2291–5. [7] Bari D, et al. IRPS; 2011. [8] Qing W et al. Electrochemical impedance spectroscopic analysis of dyesensitized solar cells. J Phys Chem B 2005;109:14945–53.